diff options
| author | Douglas Rumbaugh <dbr4@psu.edu> | 2023-11-06 15:18:53 -0500 |
|---|---|---|
| committer | Douglas Rumbaugh <dbr4@psu.edu> | 2023-11-06 15:18:53 -0500 |
| commit | e02742b07540dd5a9bcbb44dae14856bf10955ed (patch) | |
| tree | 839533401683d6ae875adbf7af2ee4c0d0f5e483 /include/shard/PGM.h | |
| parent | 0b723322a611de83872dd83b55d2e10e8886a283 (diff) | |
| download | dynamic-extension-e02742b07540dd5a9bcbb44dae14856bf10955ed.tar.gz | |
Refactoring progress
Diffstat (limited to 'include/shard/PGM.h')
| -rw-r--r-- | include/shard/PGM.h | 267 |
1 files changed, 0 insertions, 267 deletions
diff --git a/include/shard/PGM.h b/include/shard/PGM.h index 6d76376..6b66b7d 100644 --- a/include/shard/PGM.h +++ b/include/shard/PGM.h @@ -31,34 +31,6 @@ using psudb::Alias; namespace de { -template <RecordInterface R> -struct pgm_range_query_parms { - decltype(R::key) lower_bound; - decltype(R::key) upper_bound; -}; - -template <RecordInterface R> -struct PGMPointLookupParms { - decltype(R::key) target_key; -}; - -template <RecordInterface R> -class PGMRangeQuery; - -template <RecordInterface R> -class PGMPointLookup; - -template <RecordInterface R> -struct PGMState { - size_t start_idx; - size_t stop_idx; -}; - -template <RecordInterface R> -struct PGMBufferState { - size_t cutoff; -}; - template <RecordInterface R, size_t epsilon=128> class PGM { private: @@ -67,11 +39,6 @@ private: public: - - // FIXME: there has to be a better way to do this - friend class PGMRangeQuery<R>; - friend class PGMPointLookup<R>; - PGM(MutableBuffer<R>* buffer) : m_reccnt(0), m_tombstone_cnt(0) { @@ -80,8 +47,6 @@ public: m_data = (Wrapped<R>*)std::aligned_alloc(CACHELINE_SIZE, m_alloc_size); std::vector<K> keys; - //m_bf = new BloomFilter<R>(BF_FPR, buffer->get_tombstone_count(), BF_HASH_FUNCS); - size_t offset = 0; m_reccnt = 0; auto base = buffer->get_data(); @@ -110,13 +75,6 @@ public: base->header &= 3; m_data[m_reccnt++] = *base; keys.emplace_back(base->rec.key); - - /* - if (m_bf && base->is_tombstone()) { - m_tombstone_cnt++; - m_bf->insert(base->rec); - }*/ - base++; } @@ -148,8 +106,6 @@ public: } } - //m_bf = new BloomFilter<R>(BF_FPR, tombstone_count, BF_HASH_FUNCS); - m_alloc_size = (attemp_reccnt * sizeof(Wrapped<R>)) + (CACHELINE_SIZE - (attemp_reccnt * sizeof(Wrapped<R>)) % CACHELINE_SIZE); assert(m_alloc_size % CACHELINE_SIZE == 0); m_data = (Wrapped<R>*)std::aligned_alloc(CACHELINE_SIZE, m_alloc_size); @@ -172,10 +128,6 @@ public: if (!cursor.ptr->is_deleted()) { m_data[m_reccnt++] = *cursor.ptr; keys.emplace_back(cursor.ptr->rec.key); - /*if (m_bf && cursor.ptr->is_tombstone()) { - ++m_tombstone_cnt; - if (m_bf) m_bf->insert(cursor.ptr->rec); - }*/ } pq.pop(); @@ -190,15 +142,9 @@ public: ~PGM() { if (m_data) free(m_data); - //if (m_bf) delete m_bf; - } Wrapped<R> *point_lookup(const R &rec, bool filter=false) { - //if (filter && !m_bf->lookup(rec)) { - // return nullptr; - //} - size_t idx = get_lower_bound(rec.key); if (idx >= m_reccnt) { return nullptr; @@ -284,219 +230,6 @@ private: K m_max_key; K m_min_key; pgm::PGMIndex<K, epsilon> m_pgm; - //BloomFilter<R> *m_bf; -}; -template <RecordInterface R> -class PGMPointLookup { -public: - constexpr static bool EARLY_ABORT=false; - constexpr static bool SKIP_DELETE_FILTER=false; - - static void *get_query_state(PGM<R> *ts, void *parms) { - return nullptr; - } - - static void* get_buffer_query_state(MutableBuffer<R> *buffer, void *parms) { - return nullptr; - } - - static void process_query_states(void *query_parms, std::vector<void*> &shard_states, void *buff_state) { - return; - } - - static std::vector<Wrapped<R>> query(PGM<R> *ts, void *q_state, void *parms) { - std::vector<Wrapped<R>> records; - auto p = (PGMPointLookupParms<R> *) parms; - auto s = (PGMState<R> *) q_state; - - size_t idx = ts->get_lower_bound(p->target_key); - if (ts->get_record_at(idx)->rec.key == p->target_key) { - records.emplace_back(*ts->get_record_at(idx)); - } - - return records; - } - - static std::vector<Wrapped<R>> buffer_query(MutableBuffer<R> *buffer, void *state, void *parms) { - auto p = (PGMPointLookupParms<R> *) parms; - auto s = (PGMBufferState<R> *) state; - - std::vector<Wrapped<R>> records; - for (size_t i=0; i<buffer->get_record_count(); i++) { - auto rec = buffer->get_data() + i; - if (rec->rec.key == p->target_key) { - records.emplace_back(*rec); - return records; - } - } - - return records; - } - - static std::vector<R> merge(std::vector<std::vector<Wrapped<R>>> &results, void *parms) { - std::vector<R> output; - for (size_t i=0 ;i<results.size(); i++) { - if (results[i].size() > 0) { - output.emplace_back(results[i][0].rec); - return output; - } - } - - return output; - } - - static void delete_query_state(void *state) { - } - - static void delete_buffer_query_state(void *state) { - } }; - - -template <RecordInterface R> -class PGMRangeQuery { -public: - constexpr static bool EARLY_ABORT=false; - constexpr static bool SKIP_DELETE_FILTER=false; - - static void *get_query_state(PGM<R> *ts, void *parms) { - auto res = new PGMState<R>(); - auto p = (pgm_range_query_parms<R> *) parms; - - res->start_idx = ts->get_lower_bound(p->lower_bound); - res->stop_idx = ts->get_record_count(); - - return res; - } - - static void* get_buffer_query_state(MutableBuffer<R> *buffer, void *parms) { - auto res = new PGMBufferState<R>(); - res->cutoff = buffer->get_record_count(); - - return res; - } - - static void process_query_states(void *query_parms, std::vector<void*> &shard_states, void *buff_state) { - return; - } - - static std::vector<Wrapped<R>> query(PGM<R> *ts, void *q_state, void *parms) { - size_t tot = 0; - //std::vector<Wrapped<R>> records; - auto p = (pgm_range_query_parms<R> *) parms; - auto s = (PGMState<R> *) q_state; - - // if the returned index is one past the end of the - // records for the PGM, then there are not records - // in the index falling into the specified range. - if (s->start_idx == ts->get_record_count()) { - return {}; - } - - auto ptr = ts->get_record_at(s->start_idx); - - // roll the pointer forward to the first record that is - // greater than or equal to the lower bound. - while(ptr->rec.key < p->lower_bound) { - ptr++; - } - - while (ptr->rec.key <= p->upper_bound && ptr < ts->m_data + s->stop_idx) { - if (ptr->is_tombstone()) --tot; - else if (!ptr->is_deleted()) ++tot; - //records.emplace_back(*ptr); - ptr++; - } - - return {Wrapped<R>{0, {tot, 0}}}; - //return records; - } - - static std::vector<Wrapped<R>> buffer_query(MutableBuffer<R> *buffer, void *state, void *parms) { - size_t tot = 0; - auto p = (pgm_range_query_parms<R> *) parms; - auto s = (PGMBufferState<R> *) state; - - //std::vector<Wrapped<R>> records; - for (size_t i=0; i<s->cutoff; i++) { - auto rec = buffer->get_data() + i; - if (rec->rec.key >= p->lower_bound && rec->rec.key <= p->upper_bound) { - if (rec->is_tombstone()) --tot; - else if (!rec->is_deleted()) ++tot; - //records.emplace_back(*rec); - } - } - - return {Wrapped<R>{0, {tot, 0}}}; - //return records; - } - - static std::vector<R> merge(std::vector<std::vector<Wrapped<R>>> &results, void *parms) { - /*std::vector<Cursor<Wrapped<R>>> cursors; - cursors.reserve(results.size()); - - PriorityQueue<Wrapped<R>> pq(results.size()); - size_t total = 0; - size_t tmp_n = results.size(); - - - for (size_t i = 0; i < tmp_n; ++i) - if (results[i].size() > 0){ - auto base = results[i].data(); - cursors.emplace_back(Cursor{base, base + results[i].size(), 0, results[i].size()}); - assert(i == cursors.size() - 1); - total += results[i].size(); - pq.push(cursors[i].ptr, tmp_n - i - 1); - } else { - cursors.emplace_back(Cursor<Wrapped<R>>{nullptr, nullptr, 0, 0}); - } - - if (total == 0) { - return std::vector<R>(); - } - - std::vector<R> output; - output.reserve(total); - - while (pq.size()) { - auto now = pq.peek(); - auto next = pq.size() > 1 ? pq.peek(1) : queue_record<Wrapped<R>>{nullptr, 0}; - if (!now.data->is_tombstone() && next.data != nullptr && - now.data->rec == next.data->rec && next.data->is_tombstone()) { - - pq.pop(); pq.pop(); - auto& cursor1 = cursors[tmp_n - now.version - 1]; - auto& cursor2 = cursors[tmp_n - next.version - 1]; - if (advance_cursor<Wrapped<R>>(cursor1)) pq.push(cursor1.ptr, now.version); - if (advance_cursor<Wrapped<R>>(cursor2)) pq.push(cursor2.ptr, next.version); - } else { - auto& cursor = cursors[tmp_n - now.version - 1]; - if (!now.data->is_tombstone()) output.push_back(cursor.ptr->rec); - pq.pop(); - - if (advance_cursor<Wrapped<R>>(cursor)) pq.push(cursor.ptr, now.version); - } - }*/ - - size_t tot = 0; - for (auto& result: results) - if (result.size() > 0) tot += result[0].rec.key; - - return {{tot, 0}}; - } - - static void delete_query_state(void *state) { - auto s = (PGMState<R> *) state; - delete s; - } - - static void delete_buffer_query_state(void *state) { - auto s = (PGMBufferState<R> *) state; - delete s; - } -}; - -; - } |